-
EDSR PAMS 구현 model/edsr.py
-
quant weight 구현 ./model/quant_ops.py/quant_weight/init() ./model/quant_ops.py/quant_weight/forward()
-
quant activation 구현 ./model/quant_ops.py/pams_quant_act/init() ./model/quant_ops.py/pams_quant_act/forward()
-
Qconv 구현 ./model/quant_ops.py/QuantConv2d/forward()
-
SKT loss 구현 ./utils/common.py/at_loss()
#for test " sh test.sh"
#for train " sh train.sh"
#dataset 위치 ../dataset -/benchmark -Urban100 -DIV2K -/DIV2K_train_LR_bicubic -/test2k -/HR -/LR -/test4k -/HR -/LR
목표 성능 Urban100: 26.02 test2k: 27.59 test4k: 28.99